!12842 Add grad op for gpu dynamic shape testing op

From: @peilin-wang
Reviewed-by: 
Signed-off-by:

mindspore/ops/_grad/grad_other_ops.py

@@ -1,4 +1,4 @@
-# Copyright 2020 Huawei Technologies Co., Ltd
+# Copyright 2020-2021 Huawei Technologies Co., Ltd
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -83,3 +83,10 @@ def get_bprop_sync_batch_norm(self):
         dbias = out[2]
         return dx, dscale, dbias, zeros_like(mean), zeros_like(variance)
     return bprop
+
+
+@bprop_getters.register(inner.GpuConvertToDynamicShape)
+def get_bprop_gpu_convert_to_dynamic_shape(self):
+    def bprop(x, out, dout):
+        return (dout,)
+    return bprop

tests/st/ops/gpu/test_gpu_convert_to_dynamic_shape_grad_op.py

@@ -0,0 +1,56 @@
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+
+import numpy as np
+
+import mindspore.context as context
+import mindspore.nn as nn
+from mindspore import Tensor
+from mindspore.ops import composite as C
+from mindspore.ops.operations import _inner_ops as inner
+
+def test_gpu_convert_to_dynamic_shape_grad():
+    context.set_context(mode=context.GRAPH_MODE, device_target="GPU")
+
+    class Net(nn.Cell):
+        def __init__(self):
+            super(Net, self).__init__()
+            self.op = inner.GpuConvertToDynamicShape()
+
+        def construct(self, x1):
+            return self.op(x1)
+
+
+    class GradNet(nn.Cell):
+        def __init__(self, network):
+            super(GradNet, self).__init__()
+            self.grad = C.GradOperation(get_all=True, sens_param=True)
+            self.network = network
+
+        def construct(self, x1, dy):
+            return self.grad(self.network)(x1, dy)
+
+    net = Net()
+    grad_net = GradNet(net)
+
+    x1 = Tensor(np.array([1.4, -1.2, 2.5, -3.23, -4.12, 5.53]).astype(np.float32))
+    dy = Tensor(np.array([0.10, 0.11, 0.22, 0.33, 0.44, 0.155]).astype(np.float32))
+    out = grad_net(x1, dy)
+    np.testing.assert_allclose(out[0].asnumpy(), dy.asnumpy(), rtol=1e-6)
+
+    x1 = Tensor(np.array([[4.4, -6.2], [22.5, 13.23], [293, 2.22]]).astype(np.float32))
+    dy = Tensor(np.array([[0.001, 0.21], [0.22, 0.663], [0.422, 0.2]]).astype(np.float32))
+    out = grad_net(x1, dy)
+    np.testing.assert_allclose(out[0].asnumpy(), dy.asnumpy(), rtol=1e-6)